Abstract
Objectives
To identify imaging features that differentiate malignant lesions from non-malignant conditions associated with incidental uptake in the esophagus and gastroesophageal junction (GEJ) on 18F-FDG PET/CT.
Methods
Patients were included if they (1) underwent whole-body 18F-FDG PET/CT between January 2009 and July 2023, with reports documenting incidental uptake in the esophagus or GEJ and recommending upper endoscopy; and (2) had upper endoscopy within one year of the PET scan. Patients with known esophageal or GEJ cancer were excluded. 18F-FDG PET/CT images were assessed for uptake location, pattern (diffuse, segmental, or focal), eccentricity (circumferential or eccentric), maximal and mean standardized uptake values (SUVs), and CT findings (normal, wall thickening, or mass). Patients were categorized into benign conditions, Barrett’s esophagus, and malignancies based on endoscopic and pathological diagnoses.
Results
A total of 106 patients were included, comprising 82 cases of benign conditions, 11 Barrett’s esophagus, and 13 malignancies. 18F-FDG PET/CT features were not significantly different between Barrett’s esophagus and benign conditions. Compared with the Barrett’s esophagus and benign conditions, malignant lesions showed a higher frequency of isolated mid-esophageal uptake (2.2% vs. 30.8%, P = 0.048), eccentric uptake (2.2% vs. 38.5%, P < 0.001), and wall thickening or mass on CT (15.1% vs. 92.4%, P < 0.001). Lesion SUVmax/liver SUVmean demonstrated excellent diagnostic performance in predicting malignancies (area under the curve 0.94). The presence of at least one of the following five features yielded 100% sensitivity, 65.6% specificity, and 28.9% positive predictive value for malignancies: focal uptake, eccentric uptake, isolated mid-esophageal uptake, lesion SUVmax/liver SUVmean ≥ 2.8, and wall thickening or mass on CT. The presence of at least two of these features increased specificity to 94.6% while maintaining 100% sensitivity.
Conclusion
Any of the five incidental 18F-FDG PET/CT imaging features in the esophagus or GEJ warrant endoscopy.
Keywords: Esophagus, Gastroesophageal junction, 18F-FDG PET/CT, Incidental uptake, Endoscopy
Introduction
Esophageal cancer is the sixth leading cause of cancer-related deaths worldwide, with an estimated 16,130 deaths in the United States in 2024 [1, 2]. The overall prognosis for esophageal and gastroesophageal junction (GEJ) cancers remains poor, with a 5-year survival rate of 22% in the United States [3]. Due to its insidious onset and absence of early or specific symptoms, a substantial percentage of patients present with advanced-stage disease [3, 4].
Positron emission tomography/computed tomography (PET/CT) using 2-fluorine 18-fluoro-2-deoxy-d-glucose (18F-FDG) is a valuable tool for staging malignancies and detecting incidental cancers [5], frequently altering clinical decision-making [6]. However, 18F-FDG uptake in the esophagus can often be nonspecific. For example, previous studies have shown that elevated 18F-FDG uptake without morphological abnormalities in the esophagus and head and neck regions does not necessarily indicate malignancy [7, 8]. On the other hand, unnecessary workups impose additional medical, economic, and psychological burdens, and further diagnostic procedures such as endoscopy are not without risk.
Incidental esophageal uptake is commonly encountered on 18F-FDG-PET/CT, with studies reporting an incidence of approximately 1.5% [9]. Unlike focal lower gastrointestinal (GI) tract uptake, for which sufficient evidence demonstrates that a substantial proportion (50–65%) represent cancerous or precancerous lesions, thus always warranting evaluation with colonoscopy [10, 11], the percentage of malignancy associated with esophageal and GEJ incidental uptake is lower (0–30%) [7, 12, 15], and evidence guiding further management remains insufficient [16]. Previous research suggested the usage of imaging characteristics to select patients who require endoscopic evaluation for incidental upper GI uptake [7]. In clinical practice, radiology reports are important in triaging such patients [17].
Earlier studies, including a pioneering study by Roedl et al. [18] and a cross-sectional study by Stagg et al. involving diagnosed esophageal cancers [19], demonstrated that imaging features and maximum standardized uptake values (SUVmax) on 18F-FDG-PET/CT could aid in distinguishing malignant from nonmalignant etiologies. However, these studies did not specifically address incidental esophageal and GEJ uptake, and the diagnostic performance of imaging features remains unclear for subsequent management.
We hypothesized that 18F-FDG-PET/CT imaging features can help distinguish malignant from nonmalignant incidental esophageal or GEJ uptake, thereby informing management decisions, including the need for upper endoscopy. This study aims to assess the diagnostic performance of these imaging features in a real-world cohort of patients.
Methods
Patients
This retrospective single-center study was approved by the institutional review board. Informed consent was waived. Patients were included if they (1) underwent whole-body 18F-FDG PET/CT between January 2009 and July 2023, with reports documenting incidental 18F-FDG uptake in the esophagus or GEJ and recommending upper endoscopy for further evaluation; and (2) had upper endoscopy within one year of the PET scan. Patients with known esophageal or GEJ cancer were excluded. Upper endoscopic findings and pathological diagnoses from biopsy specimens were used as the gold standard. Patients were categorized into three groups: normal/benign conditions (such as esophagitis), Barrett’s esophagus, and malignancies. Benign conditions and Barrett’s esophagus were collectively defined as the non-malignant group.
Relevant clinical history was obtained through review of patients’ medical records, including prior diagnosis of gastroesophageal reflux disease (GERD), pertinent surgical history, and documentation of GERD-related symptoms and medication use at the time of PET/CT. The presence of a hiatal hernia on upper endoscopy was recorded, if available. A hiatal hernia was classified as small (≤ 5 cm) or large (> 5 cm) based on endoscopic measurements when available, or otherwise, the CT component of PET/CT [20].
PET/CT acquisition and interpretation
18F-FDG PET/CT imaging was performed using a standard protocol in Department of Radiology [21]. Patients fasted for 4–6 h prior to radiotracer injection, and serum glucose levels were confirmed to be below 200 mg/dL. The administered dose of 18F-FDG was approximately 370 MBq for patients with a body mass index < 45 kg/m2 and 555 MBq for those with a body mass index ≥ 45 kg/m2. Whole-body PET imaging was acquired 60–80 min post-injection, using 6–8 bed positions with an acquisition time of 3–5 min per bed position, depending on body mass index. All PET images were reconstructed by implementing the time-of-flight feature, and these reconstructed image series were used for analysis. A low-dose CT was acquired for attenuation correction, and a diagnostic CT was concurrently performed if requested by the ordering clinicians. Images were reviewed and analyzed on Visage (Pro Medicus Ltd). All images were reviewed by two board-certified nuclear radiologists with over five years of experience in PET/CT interpretation, who were blinded to the endoscopic and pathological diagnoses.
Incidental esophageal and GEJ uptake was qualitatively evaluated for uptake location, pattern, and eccentricity. Uptake location was categorized as proximal, mid, or distal esophagus, GEJ, or a combination of these sites. Uptake involving the entire esophagus, with or without diffuse gastric involvement, was documented separately. Uptake pattern was classified as diffuse, segmental, or focal. Diffuse uptake was defined as involving more than one-third of the esophageal length [18]. Segmental uptake involved less than one-third of the esophagus and was not in a rounded or oval configuration (Fig. 1). Focal uptake was defined as uptake confined to a rounded or oval-shaped area (Figs. 2 and 3). The uptake pattern was evaluated using both cross-sectional images and whole-body maximum intensity projection images. The latter provides an improved appreciation of the three-dimensional configuration of uptake and mitigates the impact of varying slice orientations on thin-cut cross-sectional imaging. Eccentricity was described as concentric (symmetric to the lumen or involving the entire wall) or eccentric (asymmetric to the lumen or involving a portion of the wall) [18]. Concurrent diagnostic CT (when available) or localizing CT was used to assess anatomic correlation visually, including wall thickening or mass.
Fig. 1.
A 68-year-old man underwent 18 F-FDG PET/CT for evaluation of pulmonary nodules. PET (A) and PET/CT fused image (B) demonstrated incidental increased uptake at the gastroesophageal junction (GEJ) in a segmental and concentric pattern (red arrow), with a lesion SUVmax/liver SUVmean ratio of 3.2. Localizing CT images (C) showed associated wall thickening (red arrow). The patient subsequently underwent endoscopic ultrasound, which revealed a partially circumferential GEJ mass involving approximately two-thirds of the lumen circumference (white arrowhead in D). Biopsy confirmed poorly differentiated adenocarcinoma
Fig. 2.
A 77-year-old woman with a history of lung cancer underwent 18 F-FDG PET/CT for evaluation of enlarging lung nodules. Axial PET (A), PET/CT fused image image (B), and coronal maximum intensity projection PET image (C) demonstrated isolated mid-esophageal uptake in a focal, apparent concentric pattern (red arrow), with a lesion SUVmax/liver SUVmean ratio of 2.0. No associated esophageal wall thickening was noted on the concurrent diagnostic CT (not shown). Upper endoscopy revealed a fungating mass in the middle third of the esophagus (yellow arrow in D). Biopsy confirmed invasive squamous cell carcinoma
Fig. 3.
Example of a false positive case. A 75-year-old man underwent 18 F-FDG PET/CT for evaluation of possible underlying malignancy in the setting of membranous nephropathy. Axial PET (A), PET/CT fused image (B), and coronal maximum intensity projection PET image (C) demonstrated increased distal esophageal uptake in a focal, concentric pattern (red arrow), with a lesion SUVmax/liver SUVmean ratio of 2.2. Corresponding CT images (D) showed esophageal wall thickening (red arrow). Upper endoscopy was recommended and revealed Los Angeles Grade B esophagitis
For quantitative assessment, the SUVmax and mean standardized uptake value (SUVmean) of the incidental area of uptake were measured. The regions of interest were drawn manually. The SUVmean of the normal background liver parenchyma was measured in the right hepatic lobe, and the ratio of lesion SUVmax to liver SUVmean was calculated.
Statistical analysis
Statistical analysis was conducted using R (version 4.4.2; © R Foundation for Statistical Computing). Continuous variables were summarized as medians with interquartile ranges (IQR). The Kruskal–Wallis test was used to compare continuous variables across the three groups (benign conditions, Barrett’s esophagus, and malignancies), while the Mann–Whitney U test was used for pairwise comparisons. Categorical variables were compared using the chi-square test or Fisher’s exact test, as appropriate. When a significant difference was observed, standardized residual standardized residuals were examined to identify contributing cells. Diagnostic performance metrics (sensitivity, specificity, and positive predictive value [PPV]) were calculated with 95% confidence intervals (CI). Receiver operating characteristic (ROC) analysis was performed for SUV measurements, and optimal cutoffs were determined using the minimum distance method. Precision was evaluated relative to a clinically important difference of 0.5 in lesion SUVmax/liver SUVmean between the non-malignant and malignant groups. A P value of < 0.05 was considered statistically significant.
Results
Patient characteristics
A total of 106 patients were included in this study (Table 1), including 51 women and 55 men, with a median age of 66.5 years (IQR, 57.5–74.0 years). Of these, 96 (90.6%) were White, and 95 (89.6%) were non-Hispanic. Eighty-two patients (77.4%) were categorized as having benign conditions, including 34 with normal upper endoscopy and 48 with esophagitis. Eleven patients (10.4%) were diagnosed with Barrett’s esophagus. Thirteen patients (12.3%) were found to have esophageal or GEJ malignancies, including 6 cases of squamous cell carcinoma (SCC), 6 of adenocarcinoma, and 1 case of adenocarcinoma with focal neuroendocrine differentiation. The median interval between PET/CT and upper endoscopy was 34.5 days (IQR, 14.3–79.8 days). Compared with the non-malignant group (benign conditions and Barrett’s esophagus), patients with malignancies were significantly older (65.0 [56.0–73.0] years vs. 74.0 [68.0–79.0] years, P = 0.010). No significant sex predilection was observed.
Table 1.
Demographic and 18 F-FDG PET/CT characteristics of patients with incidental esophageal or gastroesophageal junction uptake
| All patients (n = 106) | Normal/Benign (n = 82) | Barrett’s esophagus (n = 11) | Non-malignant conditions (n = 93) | Malignancies (n = 13) |
P-value |
|||
|---|---|---|---|---|---|---|---|---|
| Three groups | Malignancies vs. Non-malignant conditions | Benign vs. Barrett’s esophagus | ||||||
|
| ||||||||
| Sex | 0.10 | 1.00 | 0.070 | |||||
| Male | 55 (51.9) | 39 (47.6) | 9 (81.8) | 48 (51.6) | 7 (53.8) | |||
| Female | 51 (48.1) | 43 (52.4) | 2 (18.2) | 45 (48.4) | 6 (46.2) | |||
| Age (years) | 66.5 (57.5–74.0) | 64.5 (55.0–72.8) | 65.0 (59.5–72.5) | 65.0 (56.0–73.0) | 74.0 (68.0–79.0) | 0.035 | 0.010 | 0.80 |
| Interval between PET/CT and endoscopy (days) | 34.5 (14.3–79.8) | 40 (15.2–92.5) | 22 (14.0–29.5) | 35 (15.0–89.0) | 28 (12.0–59.0) | 0.22 | 0.49 | 0.11 |
| Lesion SUVmax | 4.2 (3.4–5.8) | 4.0 (3.2–5.1) | 3.8 (3.4–4.7) | 4.0 (3.2–5.1) | 10.1 (6.7–20.3) | < 0.001 | < 0.001 | 0.89 |
| Lesion SUVmean | 2.5 (2.1–3.1) | 2.4 (2.0–2.9) | 2.4 (2.0–2.8) | 2.4 (2.0–2.9) | 5.1 (3.4–5.3) | < 0.001 | < 0.001 | 0.89 |
| Lesion SUVmax/Liver SUVmean | 1.9 (1.6–2.3) | 1.7 (1.5–2.2) | 1.6 (1.5–1.9) | 1.7 (1.5–2.2) | 5.5 (3.2–8.1) | < 0.001 | < 0.001 | 0.49 |
| Uptake pattern | 0.14 | 0.098 | 0.50 | |||||
| Diffuse | 23 (21.7) | 22 (26.8) | 1 (9.1) | 23 (24.7) | 0 (0) | |||
| Segmental | 70 (66.0) | 50 (61.0) | 9 (81.8) | 59 (63.4) | 11 (84.6) | |||
| Focal | 13 (12.3) | 10 (12.2) | 1 (9.1) | 11 (11.8) | 2 (15.4) | |||
| Eccentricity | 0.001 | < 0.001 | 1.00 | |||||
| Concentric | 99 (93.4) | 80 (97.6) | 11 (100.0) | 91 (97.8) | 8 (61.5) | |||
| Eccentric | 7 (6.6) | 2 (2.4) | 0 (0) | 2 (2.2) | 5 (38.5) | |||
| CT correlation | < 0.001 | < 0.001 | 0.67 | |||||
| No | 80 (75.5) | 70 (85.4) | 9 (81.8) | 79 (84.9) | 1 (7.7) | |||
| Wall thickening | 20 (18.9) | 12 (14.6) | 2 (18.2) | 14 (15.1) | 6 (46.2) | |||
| Mass | 6 (5.7) | 0 (0) | 0 (0) | 0 (0) | 6 (46.2) | |||
Non-malignant group includes normal/benign cases and Barrett’s esophagus
Categorical variables are reported as counts (percentages), and continuous variables are reported as medians with interquartile ranges
Fifty-four out of 106 patients (50.9%) had a prior diagnosis of GERD before PET/CT. At the time of PET/CT, 29 patients (27.4%) had documented GERD-related symptoms, and 44 (41.5%) had documented use of GERD medications. Hiatal hernia was identified endoscopically in 36 patients (34.0%), predominantly small in size, with one large hiatal hernia noted in a patient with Barrett’s esophagus. There were no significant differences between patients with non-malignant conditions and those with malignancies in the prevalence of hiatal hernia, prior GERD diagnosis, or GERD-related symptoms and treatment (Supplementary Table 1).
PET/CT features
When comparing the three groups (benign conditions, Barrett’s esophagus, and malignancies), significant differences were observed in all SUV measurements (lesion SUVmax, lesion SUVmean, and lesion SUVmax/liver SUVmean; all P < 0.001), uptake eccentricity (P = 0.001) and CT correlation (P < 0.001). The observed differences were mainly due to distinctions between malignancies and non-malignant conditions. No significant differences in any of the assessed imaging features were found between benign conditions and Barrett’s esophagus (Table 1). Therefore, subsequent analyses focused on imaging features that could help distinguish malignancies from non-malignant processes.
Compared with non-malignant conditions, malignancies demonstrated significantly higher lesion SUVmax (4.0 [3.2–5.1] vs. 10.1 [6.7–20.3], P < 0.001), lesion SUVmean (2.4 [2.0–2.9] vs. 5.1 [3.4–5.3], P < 0.001), and lesion SUVmax/liver SUVmean (1.7 [1.5–2.2] vs. 5.5 [3.2–8.1], P < 0.001) (Table 1). There was overlap in SUV measurements between malignancies and non-malignant conditions (Fig. 4). The mean difference in lesion SUVmax/liver SUVmean between the non-malignant and malignant groups was 4.0 (95% CI, 2.2 to 5.8), and the confidence interval excluded effects smaller than the clinically important threshold of 0.5, indicating adequate precision for comparison. In ROC analysis, all SUV measurements demonstrated excellent diagnostic performance. Lesion SUVmax achieved an area under the curve (AUC) of 0.93 (95% CI, 0.86–1.00), with an optimal cutoff value of 6.1. Lesion SUVmean yielded an AUC of 0.93 (95% CI, 0.87–0.99), with an optimal cutoff of 3.2. Lesion SUVmax/liver SUVmean, which is generally a more reproducible measure [22], demonstrated an AUC of 0.94 (95% CI, 0.89–1.00), with an optimal cutoff of 2.8 (Fig. 5).
Fig. 4.
Lesion SUVmax, lesion SUVmean, and lesion SUVmax/liver SUVmean in patients with benign conditions and Barrett’s esophagus versus malignancies. The black horizontal bars in the scatter plot indicate the median. The dotted lines indicate optimal cutoff values
Fig. 5.
Receiver operating characteristic (ROC) curves for lesion SUVmax, lesion SUVmean, and lesion SUVmax/liver SUVmean in the prediction of malignant lesions. AUC area under the curve
For qualitative imaging features, compared with non-malignant processes, malignant lesions were more likely to demonstrate eccentric uptake (2.2% vs. 38.5%, P < 0.001) and associated wall thickening or mass on the CT (15.1% vs. 92.4%, P < 0.001). Interestingly, no significant difference in uptake pattern was observed between non-malignant and malignant lesions (P = 0.098) (Table 1). Regarding the location of 18F-FDG uptake, none of the patients with diffuse esophageal involvement without superimposed focality were found to have malignancies. Among patients with non-diffuse uptake, a significant difference in uptake location was observed between those with non-malignant conditions and those with malignancies (P = 0.048). Standardized residual analysis indicated that this difference was primarily driven by the proportion of patients with isolated mid-esophageal uptake (2.2% in the non-malignant group vs. 30.8% in the malignant group; absolute standardized residual, 3.65) (Table 2).
Table 2.
Location of incidental esophageal or gastroesophageal junction 18 F-FDG uptake
| Location | All patients (n = 106) | Benign conditions and Barrett’s esophagus (n = 93) | Malignancies (n = 13) | P-value |
|---|---|---|---|---|
|
| ||||
| Non-diffuse involvement | 86 (81.1) | 73 (78.5) | 13 (100.0) | 0.048 |
| Proximal esophagus only | 4 (3.8) | 3 (3.2) | 1 (7.7) | |
| Mid esophagus only | 6 (5.7) | 2 (2.2) | 4 (30.8) | |
| Distal esophagus only | 51 (48.1) | 46 (49.5) | 5 (38.5) | |
| GE junction only | 10 (9.4) | 9 (9.7) | 1 (7.7) | |
| Mid esophagus + distal esophagus | 5 (4.7) | 4 (4.3) | 1 (7.7) | |
| Distal esophagus + GE junction | 5 (4.7) | 4 (4.3) | 1 (7.7) | |
| Distal esophagus + gastric fundus | 2 (1.9) | 2 (2.2) | 0 (0) | |
| GE junction + gastric fundus | 3 (2.8) | 3 (3.2) | 0 (0) | |
| Diffuse involvement | 20 (18.9) | 20 (21.5) | 0 (0) | NA |
| Esophagus | 18 (17.0) | 18 (19.4) | 0 (0) | |
| Esophagus and stomach | 2 (1.9) | 2 (2.2) | 0 (0) | |
The p-value is obtained from the Chi-square test within the non-diffuse involvement subpopulation
Categorical variables are reported as counts (percentages)
GE junction gastroesophageal junction, NA not applicable
Among the six patients with SCC, one case showed increased uptake in the proximal esophagus, four in the mid esophagus, and one in the distal esophagus. Among the six patients with adenocarcinoma, five showed abnormal uptake in the distal esophagus and/or GEJ, and one in the mid and distal esophagus. The adenocarcinoma case with focal neuroendocrine differentiation exhibited uptake in the distal esophagus (Supplementary Table 2).
Diagnostic performance of imaging features
Based on the previous analyses and prior studies [18, 19], we examined the diagnostic performance of five PET/CT features potentially useful in differentiating esophageal or GEJ malignancies from non-malignant processes: focal uptake, eccentric uptake, isolated mid-esophageal uptake, lesion SUVmax/liver SUVmean ≥ 2.8, and wall thickening or mass on CT. All five features demonstrated a high specificity in predicting malignancy, ranging from 85% to 98% (Table 3). Except for focal uptake, these features had PPVs ranging from 46% to 71%. Focal uptake exhibited a lower, but still clinically relevant, PPV of 15% (95% CI, 4–42%). CT correlation and lesion SUVmax/liver SUVmean ≥ 2.8 also demonstrated a high sensitivity (both 92% [95% CI, 67% to 99%]), while the other three features had lower sensitivities, ranging from 15% to 38%. It should be noted that although segmental uptake was common in the malignant group (84.6%), it was frequently observed in the non-malignant group (63.4%), yielding low specificity (37%). Therefore, it was not included among the differentiating features.
Table 3.
Diagnostic performance of various imaging criteria for predicting esophageal or gastroesophageal junction malignancies
| PET/CT features | Sensitivity (%) | Specificity (%) | PPV (%) |
|---|---|---|---|
|
| |||
| Focal uptake | 15 (4–42) | 88 (80–93) | 15 (4–42) |
| Eccentric uptake | 38 (18–64) | 98 (92–99) | 71 (36–92) |
| CT correlation with wall thickening or mass | 92 (67–99) | 85 (76–91) | 46 (29–65) |
| Isolated mid-esophageal uptake | 31 (13–58) | 98 (92–99) | 67 (30–90) |
| Lesion SUVmax/liver SUVmean ≥ 2.8 | 92 (67–99) | 91 (84–96) | 60 (39–78) |
| Number of the above features present | |||
| At least 1 feature | 100 (77–100) | 66 (55–74) | 29 (18–43) |
| At least 2 features | 100 (77–100) | 95 (88–98) | 72 (49–88) |
| At least 3 features | 54 (29–77) | 100 (96–100) | 100 (65–100) |
| At least 4 features | 8 (1–33) | 100 (96–100) | 100 (21–100) |
PPV positive predictive value, NPV negative predictive value
We subsequently assessed the diagnostic performance of combining these imaging features (Table 3). A new composite criterion, defined as the presence of at least one of the five features, yielded a sensitivity of 100% (95% CI, 77% to 100%), specificity of 66% (95% CI, 55% to 74%), and a PPV of 29% (95% CI, 18% to 43%) for predicting malignancies. This criterion was met by all 13 malignant cases, as well as by 5 cases of Barrett’s esophagus, 17 cases of esophagitis, and 5 normal cases. Requiring the presence of at least two features increased specificity to 95% (95% CI, 88% to 98%) and PPV to 72% (95% CI, 49% to 88%), while maintaining a sensitivity of 100% (95% CI, 77% to 100%). All 13 malignancies met this stricter criterion, along with 3 esophagitis cases and 2 normal cases. To inform clinical decision-making in more complex scenarios, detailed data on the PPVs of various combinations of two imaging features are provided in Supplementary Table 3.
The presence of the five PET/CT features in the 13 malignancy cases is summarized in Table 4. All cases with incidentally identified malignancies demonstrated at least two of the five features; six cases exhibited three features, and one case demonstrated all five. Of note, all malignant cases with isolated mid-esophageal uptake were SCC. Examples of true-positive and false-positive cases are shown in Figs. 1, 2 and 3.
Table 4.
Summary of the 18 F-FDG PET/CT characteristics of the 13 malignancy cases
| Case | Pathologic diagnosis | Focal uptake | Eccentric uptake | CT correlation with wall thickening or mass | Isolated mid esophagus uptake | Lesion SUVmax/liver SUVmean ≥ 2.8 | Number of features identified |
|---|---|---|---|---|---|---|---|
|
| |||||||
| 1 | Adenocarcinoma | − | − | + | − | + | 2 |
| 2 | Adenocarcinoma | − | − | + | − | + | 2 |
| 3 | Adenocarcinoma | − | + | + | − | + | 3 |
| 4 | Adenocarcinoma | − | − | + | − | + | 2 |
| 5 | Adenocarcinoma | − | + | + | − | + | 3 |
| 6 | Adenocarcinoma | − | − | + | − | + | 2 |
| 7 | Adenocarcinoma with focal neuroendocrine differentiation | − | − | + | − | + | 2 |
| 8 | SCC | + | + | + | + | + | 5 |
| 9 | SCC | − | + | + | − | + | 3 |
| 10 | SCC | − | + | + | − | + | 3 |
| 11 | SCC | + | − | − | + | − | 2 |
| 12 | SCC | − | − | + | + | + | 3 |
| 13 | SCC | − | − | + | + | + | 3 |
SCC squamous cell carcinoma
Discussion
Our study is the first to investigate 18F-FDG PET/CT features that could help distinguish malignancies from non-malignant conditions in cases with incidental esophageal or GEJ uptake, intending to improve patient triage. This represents an extension of the pioneering case-control study by Roedl et al. [18] on a similar topic; in contrast, we employed a cross-sectional design focused on incidental uptake in a real-world clinical setting, allowing us to report PPVs. We found that a combination of five imaging features can guide the management of incidental esophageal or GEJ uptake on 18F-FDG PET/CT.
In our cohort, the proportion of malignancies among cases with incidental uptake was 12%, consistent with previously reported ranges (0–30%) [7, 12, 15]. Approximately one-third of patients had a normal upper endoscopy. While this may be partially explained by the time gap between 18F-PET/CT and endoscopy, prior studies on GERD have shown that half of patients with GERD may exhibit increased esophageal uptake despite normal endoscopy, potentially reflecting microscopic inflammatory changes [9].
Barrett’s esophagus and normal/benign conditions did not show significant differences across any of the assessed PET/CT features. This is pathophysiologically plausible - compared with esophagitis without intestinal metaplasia, the areas of intestinal metaplasia characteristic of Barrett’s esophagus may not produce distinguishable differences on PET/CT imaging. A similar observation has been reported in colonic adenomas, another type of precancerous lesion, where SUVmax fails to distinguish them from benign lesions [23].
SUV measurements demonstrated strong diagnostic value for esophageal or GEJ malignant tumors, consistent with previous studies [15, 19]. Stagg et al. reported an AUC of 0.97 for SUVmax and proposed a threshold of SUVmax ≥ 3.5 for predicting esophageal malignancy [19]. Hosni et al. suggested a cutoff of SUVmax 9.2 for identifying cancer or pre-cancerous lesions in both the upper and lower gastrointestinal tracts [15]. Our optimal SUVmax cutoff of 6.1 falls within this range. On the other hand, it is important to note that SUV values in malignancies can vary widely [24], and there is notable overlap between malignant and non-malignant conditions. For reference, Heusner et al. reported a mean SUVmax of 3.3 (range, 1.0–8.4) in the normal GEJ [14], while Jo et al. observed SUVmax values ranging from 2.3 to 8.4 in patients with reflux esophagitis [25].
For qualitative PET/CT features, our study reaffirmed the diagnostic value of eccentric uptake [18]. This is a specific but not sensitive feature. The limited sensitivity may be attributed to several factors: some malignant lesions may inherently exhibit circumferential involvement; due to the spatial resolution limits of PET, even lesions that appear eccentric on endoscopy may not show clearly eccentric uptake on imaging; and intraluminal fungating masses may also result in apparent concentric uptake pattern in a collapsed esophagus.
Interestingly, unlike the previous study by Roedl et al. [18], focal uptake did not show a significant difference between malignant and non-malignant conditions. This is primarily due to the lower proportion of focal uptake among malignant cases in our cohort. Additionally, Roedl et al. applied a broader definition of focal uptake - lesions less than 3 cm in length with SUV > 3 - while we defined focal uptake as having a rounded or oval configuration, which more closely reflects our practical interpretation in clinical settings. Nevertheless, focal uptake remains a specific feature with moderately high specificity.
We found that isolated mid-esophageal uptake was more common in malignancies and represented a highly specific imaging feature. This can be explained by the underlying pathophysiology: abnormalities related to reflux esophagitis and Barrett’s esophagus typically originate in the distal esophagus and extend proximally. As such, isolated mid-esophageal uptake without distal involvement is uncommon in these non-malignant conditions. Moreover, this feature may also provide insight into the pathological type of malignancy. In our cohort, all malignant cases with isolated mid-esophageal uptake were SCC. In contrast to adenocarcinoma - which is associated with GERD and Barrett’s esophagus and thus more often arises in the distal esophagus - SCC is not typically linked to these conditions and more commonly occurs in the proximal to mid esophagus [26, 27]. On the other hand, all cases of diffuse uptake without superimposed focality in our study were non-malignant, a finding that has been consistently reported in prior studies [13, 28].
We propose that the presence of any one of the following features should prompt further evaluation with upper endoscopy: focal uptake, lesion SUVmax/liver SUVmean ≥ 2.8, eccentric uptake, isolated mid-esophageal uptake, or wall thickening or mass on CT. Although applying a stricter criterion of requiring at least two features increased specificity while maintaining 100% sensitivity, further evaluation is warranted even when one feature is present, given the serious consequences of missing a malignancy. This less strict criterion may reduce the risk of overfitting and, in our dataset, allowed for the detection of 5 out of 11 cases of Barrett’s esophagus. It is worth noting that identifying esophagitis also carries clinical value, especially given the difficulty of distinguishing concurrent Barrett’s esophagus on PET/CT. In such cases, reporting inflammatory uptake may increase clinical vigilance, prompt treatment, and facilitate appropriate follow-up. Importantly, the presence or absence of hiatal hernia, prior GERD diagnosis, GERD-related symptoms, and anti-reflux medication use should not alter the recommendation for upper endoscopy based on imaging features.
This study has several limitations. First, this was a single-center study with a relatively small sample size, and the majority of participants were White. Larger multicenter studies encompassing more diverse racial and ethnic populations are warranted to enhance the generalizability of our findings. Second, we allowed a relatively long interval (up to one year) between PET/CT and upper endoscopy to increase case inclusion. This may have limited the accuracy of endoscopic findings in reflecting the physiological status at the time of PET/CT, especially for conditions like esophagitis, which may fluctuate over time. However, for malignancies, PET/CT was performed before endoscopy (median interval of approximately 1 month, maximum 6 months). Therefore, the PET/CT findings likely represent earlier-stage tumors, which typically exhibit more subtle imaging features. Third, multiple PET/CT scanners were used within the radiology department during the study period, which could have introduced minor variations in imaging technique.
In summary, a combination of five imaging features can guide further management of incidental esophageal or GEJ uptake on 18F-FDG PET/CT. The presence of any one of the following features should prompt upper endoscopy: focal uptake, lesion SUVmax/liver SUVmean ≥ 2.8, eccentric uptake, isolated mid-esophageal uptake, and wall thickening or mass on CT. 18F-FDG PET/CT cannot reliably distinguish Barrett’s esophagus from other non-malignant processes.
Supplementary Material
Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00261-025-05257-4.
Acknowledgements
We thank Yitong Liu, for histological review of a subset of patients.
Footnotes
Competing interests SAE is supported by NCI K08-CA259626 and has received research support from Sofie Biosciences, Novartis, and Telix Pharmaceuticals, and is a consultant for Telix and Sofie. SAE is an assistant editor of Abdominal Radiology. UM is a co-founder, shareholder, and consultant (Scientific Advisory Board) of CytoSite BioPharma. PH is supported by NCI K08-CA249047 and is a consultant for Telix and serves on the advisory board of Novartis. TN is supported in part by the U.S. National Institutes of Health (R21EB036323, R01DK097112, R01CA293888, R01EB034692), the U.S. Department of Defense (W81XWH-22–1-0061), the Lee Family Foundation, and has received research support from Lantheus and Bayer. SJK has served in a consultant/advisory role for Bristol Myers Squibb, Merck, Amgen, Astellas, Daiichi-Sankyo, Novartis, AstraZeneca, Mersana, I-Mab, Gilead, BeiGene, Elevation Oncology, and EsoBiotec. SJK is supported by the DFCI/HCC GI SPORE (P50CA127003). SJK reports stock/equity in Turning Point Therapeutics and Nuvalent. MRS has served in a consultant/advisory role for Bristol Myers Squibb, Astellas, Sedgewick Claims Management, Bayer, Encapsulate, and UpToDate. Other authors declare no conflicts of interest.
Data availability
Data is provided within the manuscript or supplementary information files.
References
- 1.Ajani JA, Barthel JS, Bentrem DJ et al. (2011) Esophageal and Esophagogastric Junction Cancers. J Natl Compr Canc Netw 9:830–887. 10.6004/jnccn.2011.0072 [DOI] [PubMed] [Google Scholar]
- 2.Siegel RL, Giaquinto AN, Jemal A (2024) Cancer statistics, 2024. CA A Cancer J Clinicians 74:12–49. 10.3322/caac.21820 [DOI] [PubMed] [Google Scholar]
- 3.Yang H, Wang F, Hallemeier CL et al. (2024) Oesophageal cancer. The Lancet 404:1991–2005. 10.1016/S0140-6736(24)02226-8 [DOI] [PubMed] [Google Scholar]
- 4.Deboever N, Jones CM, Yamashita K, et al. (2024) Advances in diagnosis and management of cancer of the esophagus. BMJ e074962. 10.1136/bmj-2023-074962 [DOI] [PubMed] [Google Scholar]
- 5.Sone Y, Sobajima A, Kawachi T et al. (2014) Ability of 18-flude-oxyglucose positron emission tomography/CT to detect incidental cancer. BJR 87:20140030. 10.1259/bjr.20140030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Winkens T, Rudakoff W, Rauchfuss F et al. (2021) FDG PET/CT to Detect Incidental Findings in Patients With Hepatocellular Carcinoma—Additional Benefit for Patients Considered for Liver Transplantation? Clin Nucl Med 46:532–539. 10.1097/RLU.0000000000003576 [DOI] [PubMed] [Google Scholar]
- 7.Correia Gomes L, Fraga D, Lage P, et al. (2024) Incidental Uptake in Gastrointestinal Tract on 18F-FDG-PET/CT: Is It Worth to Investigate? A Study with 371 Patients. GE Port J Gastroenterol 1–11. 10.1159/000541209 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Heusner TA, Hahn S, Hamami ME et al. (2009) Incidental head and neck 18F-FDG uptake on PET/CT without corresponding morphological lesion: early predictor of cancer development? Eur J Nucl Med Mol Imaging 36:1397–1406. 10.1007/s00259-009-1113-1 [DOI] [PubMed] [Google Scholar]
- 9.Kamboj AK, Agarwal S, Carroll EF et al. (2022) Association of Incidental Positron Emission Tomography Uptake in the Esophagus to Gastroesophageal Reflux Disease. Clinical Gastroenterology and Hepatology 20:2644–2646.e1. 10.1016/j.cgh.2021.08.044 [DOI] [PubMed] [Google Scholar]
- 10.Treglia G, Calcagni ML, Rufini V et al. (2012) Clinical significance of incidental focal colorectal18 F-fluorodeoxyglucose uptake: our experience and a review of the literature. Colorectal Disease 14:174–180. 10.1111/j.1463-1318.2011.02588.x [DOI] [PubMed] [Google Scholar]
- 11.Lee C, Koh S-J, Kim JW et al. (2013) Incidental Colonic 18F-Fluorodeoxyglucose Uptake: Do We Need Colonoscopy for Patients with Focal Uptake Confined to the Left-Sided Colon? Dig Dis Sci 58:229–235. 10.1007/s10620-012-2333-y [DOI] [PubMed] [Google Scholar]
- 12.Chopra A, Ford A, De Noronha R, Matthews S (2012) Incidental findings on positron emission tomography/CT scans performed in the investigation of lung cancer. BJR 85:e229–e237. 10.1259/bjr/60606623 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Vella-Boucaud J, Papathanassiou D, Bouche O et al. (2015) Incidental gastrointestinal 18F-Fluorodeoxyglucose uptake associated with lung cancer. BMC Pulm Med 15:152. 10.1186/s12890-015-0152-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Heusner TA, Hahn S, Hamami ME et al. (2009) Gastrointestinal 18F-FDG accumulation on PET without a corresponding CT abnormality is not an early indicator of cancer development. Eur Radiol 19:2171–2179. 10.1007/s00330-009-1405-7 [DOI] [PubMed] [Google Scholar]
- 15.Hosni MN, Kassas M, Itani MI et al. (2023) The Clinical Significance of Incidental GIT Uptake on PET/CT: Radiologic, Endoscopic, and Pathologic Correlation. Diagnostics 13:1297. 10.3390/diagnostics13071297 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Pencharz D, Nathan M, Wagner TL (2018) Evidence-based management of incidental focal uptake of fluorodeoxyglucose on PET-CT. The British Journal of Radiology 91:20170774. 10.1259/bjr.20170774 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wang G, Lau EWF, Shakher R et al. (2007) How do oncologists deal with incidental abnormalities on whole-body fluorine-18 fluorodeoxyglucose PET/CT? Cancer 109:117–124. 10.1002/cncr.22370 [DOI] [PubMed] [Google Scholar]
- 18.Roedl JB, Colen RR, King K et al. (2008) Visual PET/CT Scoring for Nonspecific18 F-FDG Uptake in the Differentiation of Early Malignant and Benign Esophageal Lesions. American Journal of Roentgenology 191:515–521. 10.2214/AJR.07.3320 [DOI] [PubMed] [Google Scholar]
- 19.Stagg J, Farukhi I, Lazaga F et al. (2015) Significance of 18F-Fluorodeoxyglucose Uptake at the Gastroesophageal Junction: Comparison of PET to Esophagogastroduodenoscopy. Dig Dis Sci 60:1335–1342. 10.1007/s10620-014-3456-0 [DOI] [PubMed] [Google Scholar]
- 20.Abu-Freha N, Guterman R, Elhayany R et al. (2023) Hiatal hernia: risk factors, and clinical and endoscopic aspects in gastroscopy. Gastroenterology Report 12:goae086. 10.1093/gastro/goae086 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Esfahani SA, Heidari P, Halpern EF, et al. Baseline total lesion glycolysis measured with 18F-FDG PET/CT as a predictor of progression-free survival in diffuse large B-cell lymphoma: a pilot study [PMC free article] [PubMed] [Google Scholar]
- 22.Groshar D, Bernstine H, Goldberg N et al. (2017) Reproducibility and repeatability of same-day two sequential FDG PET/MR and PET/CT. Cancer Imaging 17:11. 10.1186/s40644-017-0113-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Van Hoeij FB, Keijsers RGM, Loffeld BCAJ et al. (2015) Incidental colonic focal FDG uptake on PET/CT: can the maximum standardized uptake value (SUVmax) guide us in the timing of colonoscopy? Eur J Nucl Med Mol Imaging 42:66–71. 10.1007/s00259-014-2887-3 [DOI] [PubMed] [Google Scholar]
- 24.Özaslan E, Kiziltepe M, Addulrezzak Ü et al. (2021) Is SUVmax of 18F-FDG PET/CT Predictive Factor for Malignancy in Gastrointestinal Tract? Nigerian Journal of Clinical Practice 24:1217–1224. 10.4103/njcp.njcp_637_18 [DOI] [PubMed] [Google Scholar]
- 25.Jo K, Kim S, Cha J et al. (2016) A Comparison Study of Esophageal Findings on 18F-FDG PET/CT and Esophagogastroduodenoscopy. Nucl Med Mol Imaging 50:123–129. 10.1007/s13139-015-0376-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Prabhakar HB, Sahani DV, Fischman AJ et al. (2007) Bowel Hot Spots at PET-CT. RadioGraphics 27:145–159. 10.11148/rg.271065080 [DOI] [PubMed] [Google Scholar]
- 27.Rustgi AK, El-Serag HB (2014) Esophageal Carcinoma. N Engl J Med 371:2499–2509. 10.1056/NEJMra1314530 [DOI] [PubMed] [Google Scholar]
- 28.Kamel EM, Thumshirn M, Truninger K et al. (2004) Significance of Incidental 18F-FDG Accumulations in the Gastrointestinal Tract in PET/CT: Correlation with Endoscopic and Histopathologic Results. Journal of Nuclear Medicine 45:1804–1810 [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data is provided within the manuscript or supplementary information files.